Google does not penalize content because AI helped write it. It penalizes content that exists mainly to manipulate rankings or AI answers and gives readers nothing new, regardless of how that content was produced. The September 2026 spam update, which started on September 24, did not change this rule. Google's guidance on AI-generated content was last updated in December 2025, and the announcement of the update does not name any target.
Still, I believe the risk for companies that publish automated content is higher now than it was a year ago. Below I explain why, using what Google itself has published and what I see in our work at EntityRise.
What did Google change on September 24, 2026?
On September 24, 2026, Google released the September 2026 spam update, and it was the only ranking change Google announced that day. The Google Search Status Dashboard says the update applies globally and to all languages and that the rollout may take up to two weeks. It says nothing about AI content, links or any specific tactic.
Two other things happened the same day, and neither concerns AI-written content. Search Console began reporting searches made with images, such as Google Lens and Circle to Search, including in its report for generative AI features. Google's documentation changelog for that day has a single entry, about video structured data. Google's page on using generative AI content was last updated on December 10, 2025, and its spam policies page on August 28, 2026, when Google adjusted how it enforces the site reputation policy in the European Economic Area.
What caught my attention is the length of the rollout window. Every earlier spam update in 2026 finished in less than three days. This one was announced with a window of up to two weeks. Google has not explained why, so I will not guess.
| Update | Started | Duration |
|---|---|---|
| Discover update | February 5 | 21 days, 17 hours |
| Spam update | March 24 | 19 hours, 30 minutes |
| Core update | March 27 | 12 days, 4 hours |
| Core update | May 21 | 11 days, 21 hours |
| Spam update | June 24 | 2 days, 1 hour |
| Spam update | August 18 | 2 days, 16 hours |
| Spam update | September 24 | in progress, up to 2 weeks |
Does Google penalize AI-written content?
No. Google judges content by whether it helps people, not by the tool used to write it. Its policy on scaled content abuse covers large amounts of unoriginal content that adds little or no value, regardless of how that content was created.
Google's examples of scaled content abuse include:
using generative AI tools or similar tools to generate many pages without adding value for users;
scraping feeds or search results and transforming them automatically, for example by swapping synonyms or translating;
stitching together content from different pages without adding value;
creating multiple sites to hide how much content is being produced.
Google's guidance also says that generative AI can be useful for researching a topic and adding structure to original content. The line does not run between AI and humans. It runs between value and volume.
One detail matters for online stores: Google Merchant Center requires AI-generated product images to carry metadata that marks them as AI-generated, and AI-generated product titles and descriptions to be submitted separately and labeled.
Why is automated blog content a bigger risk now?
This is where I see the real danger. Many products today sell automated blog writing as a way to grow in SEO. They produce a lot of articles quickly, and each of them tends to summarize what is already online.
That is exactly what Google calls commodity content in its guide to generative AI features in Search, published in May 2026. Google contrasts it with content that brings a unique point of view or first-hand experience, and it warns against recycling what others have already said or what a generative AI model could easily produce. The same guide says that creating separate pages for every variation of a query, including the related queries AI systems run in the background, violates the scaled content abuse policy when the goal is to manipulate rankings or AI answers.
Since May 15, 2026, Google's spam policies have also stated openly that trying to manipulate generative AI responses in Search is spam. The AI answer is no longer a side channel with softer rules.
So in my view, a blog filled with automated articles can lower a company's position in search and in AI answers instead of raising it. I am not saying AI has no place in writing. I am saying that volume without a person who knows the subject has become a liability.
What does Google tell you to ignore?
Google's guide lists several popular "AEO" and "GEO" tactics that Google Search does not need:
llms.txt files and other special AI files or markup; Google says they neither help nor harm rankings;
splitting content into small "chunks" for AI;
rewriting content just for AI systems;
chasing inauthentic mentions around the web;
overfocusing on structured data, which helps with rich results but is not required for AI features.
We keep an llms.txt file on entityrise.ai, because other assistants may read it. For Google Search it is neutral, and Google says so openly.
Google also writes that, from its perspective, optimizing for generative AI search is still SEO. For Google Search, I agree. But Google speaks only for Google. ChatGPT, Claude, Grok and Perplexity are separate systems with their own models and sources, and Google's guide does not describe how they choose what to recommend. For how those systems differ from classic SEO, see AI visibility vs SEO.
Why can nobody guarantee visibility in AI answers?
Nobody can guarantee that AI assistants will recommend your brand, because the systems that decide keep changing. Google itself writes that a page meeting every requirement is still not guaranteed to be crawled, indexed or shown. About outside tools it is even more direct: "No third-party tool has access to our internal ranking or AI systems."
Google also explains that its AI features are built on its core ranking and quality systems: they retrieve pages from the Search index and run related queries in the background to gather more results. In practice, every core or spam update can change which pages AI Overviews and AI Mode rely on. In 2026 alone, Google confirmed seven ranking updates in less than eight months, between February 5 and September 24.
This matches what I see in my work. What helps AI show your company today can stop working in two or three months, when Google changes its systems or a model gets a new version and starts giving more weight to other sources of truth about your company. That is why every EntityRise audit records the model version of each AI provider and the time of the test. A result is true for a specific moment, not forever. The process is public on our methodology page.
So if someone promises you 100% visibility in AI, permanently, they are not telling you the truth. They are monetizing your uncertainty. What you can do, and in my opinion should do, is stay fully compliant with the rules, measure regularly and keep working on the signals AI systems use to understand your company.
Is search moving from Google to AI?
Search is moving to AI, and a large part of that AI is Google itself. According to figures Google has published:
AI Overviews have over 2.5 billion monthly active users (Sundar Pichai, Google I/O, May 19, 2026);
AI Mode passed 1 billion monthly active users (Google I/O, May 19, 2026, repeated on Alphabet's earnings call on July 22, 2026);
Google says it sends billions of clicks to websites every week through AI features in Search (July 22, 2026);
since August 31, 2026, every site in Search Console has a dedicated report on its impressions in Google's generative AI features.
People increasingly ask for a recommendation and get a short answer instead of ten links. That is where many buying decisions start today. And because the models behind those answers keep evolving, the way they perceive your company and your products keeps changing too. That is the core of AI visibility.
Where EntityRise stands: a person behind every change
This is the part I care about most. At EntityRise, measurement is automated, because it has to be repeatable: you approve 20 buyer questions, we seal the set before the first test and run it across ChatGPT, Claude, Gemini, Grok and Perplexity, which gives up to 100 observations. Everything that changes a client's website, content or presence elsewhere on the web is done by people, and each change is verified live. In the Expert Audit, a person checks every finding before it reaches the client.
Our team has been working together since 2017, across AI, SEO, GEO and AEO. We have watched these systems change for years, and that is exactly why we do not promise permanence. We tell clients how it really works: visibility can be measured and improved, but it has to be maintained.
What I would do now
Frequently asked questions
Does Google penalize content written with AI?
No. Google evaluates whether content helps users, regardless of how it was created. Using generative AI tools to produce many pages without adding value is treated as scaled content abuse under Google's spam policies.
What did the September 2026 spam update target?
Google did not say. The Search Status Dashboard states only that the update started on September 24, 2026, applies globally and to all languages, and may take up to two weeks.
Does llms.txt help visibility in Google Search?
No. Google says its Search ignores llms.txt files, so they neither help nor harm rankings. They can still be useful for other AI systems.
Can anyone guarantee that AI assistants will recommend my company?
No. Google states that meeting every requirement does not guarantee crawling, indexing or serving, and AI models change their sources and versions over time. Visibility in AI can be measured and improved, but not guaranteed.
How can I check my visibility in Google's AI features and other assistants?
Google Search Console has a Generative AI performance report, available to all sites since August 31, 2026. For ChatGPT, Claude, Gemini, Grok and Perplexity, EntityRise measures how each assistant describes and recommends your company.
Sources
- 1. Google Search Status Dashboard: September 2026 spam update
- 2. Google Search Status Dashboard: ranking update history
- 3. Google Search Central: latest documentation updates
- 4. Google Search Central: spam policies for Google web search
- 5. Google Search Central: guidance on using generative AI content
- 6. Google Search Central: optimizing for generative AI features on Google Search
- 7. Google Search Central Blog: Search generative AI performance reports (June 3, 2026)
- 8. Google Search Central Blog: web multimodal reporting (September 24, 2026)
- 9. Google I/O 2026: Sundar Pichai's keynote
- 10. Alphabet Q2 2026 earnings call: Sundar Pichai's remarks
Kamil Kępiński · Founder of EntityRise
Author of the EntityRise AI visibility measurement methodology: 20 buyer questions, five providers, up to 100 recorded observations. He writes about how models choose the companies they recommend, and how to measure it for yourself.
Visibility can be measured and improved, but it has to be maintained.
EntityRise measures how assistants describe you, implements the fixes and remeasures after major updates.